Dear Douglas,
no I did not run in GLM-fit - I run in qdec
The conversion into z-scores was done in a statistical program (statistica) and was calculated on basis of all subjects (both groups)
- the thing that confuses me is that - if I understand it correctly from previous discussions on the mailing list - it should not be needed to demean age in order to correct for it? So in qdec I get completely different results if i use z-scores - than if I use the original raw age data. If I look at the plot of the z-scores versus raw age in qdec the distribution on the plot is exactly the same. So I would expect that no matter using raw age or z-scores would affect results the same? But the results look completely different (basically the raw-age corrected data looks wrong)
Am I doing something wrong?
thanks Olof